Blog
Responsible AI
December 22, 2023

Personal Data, Rentier Innovation, and the Global South

Can a model of personal data as an economic asset deliver AI for social good? At a talk at the Open Data Institute on AI and Infrastructures in the global south, Urvashi characterized current innovation paradigms as one of 'rentier innovation'. This paradigm is harmful and unsustainable. Listen to her talk, and see the summary notes from ODI.
Download PDF
Personal Data, Rentier Innovation, and the Global South
illustration by:

The history of current AI and Machine Learning developments was explored to highlight how data practice and governance can avoid exacerbating inequalities in the Global South. By looking at how the interests of the advertising industry have shaped the way data is generated and captured by big tech platforms, claims made about value creation in digital infrastructure were questioned. Instead it was proposed to frame big tech business models as a new form of rentier innovation: this framing of data-driven economic activities as rentiership enables a new critical perspective on the development of the Fourth Industrial Revolution. For example: how do we avoid turning personal data into the currency of our times? How do we ensure people and communities have meaningful choices in regards to how data about them is collected and used?

1 | Less data, more context

It was argued that most challenges and tensions around how to govern data infrastructure in the Global South tend to be about competing values and interests; however, it was also argued that data availability and data in itself contribute to determining these values and interests, or may even be used strategically for shaping these determinations.

Participants discussed how framing policy-making as being data-driven, and assuming that more data will always solve policy questions, can mask issues around the underlying problems that policy needs to address. For example, framing issues around malnutrition amongst poor populations as a problem that could be solved if we had access to the right data may divert attention from other issues that may be driving it, such as corruption and mismanagement. Consequently, the importance of being able to diagnose policy problems is essential for making sure there is clarity and transparency on what the data can and will be useful for, and therefore which data needs to be collected. For this reason, it was argued that a key question to consider is: what kind of policy-making is data extraction in the Global South enabling; and whether the benefits for policy-making of expanding data extraction justify this extractivism or its expansion. 

Participants also noted how over-reliance on data for policy-making shapes how we evaluate and compare the level of development of different territories or jurisdictions in the Global South in ways that may be misleading. It was observed that relying on data without understanding the context on the ground in which that data is generated may cause mismatches between the interpretation of the data by policy-makers and the material or experiential realities in the domain about which policy is being made. This raised questions about who is able to access, provide, use and interpret data for policy-making, and how that ability is intertwined with other sources of power or wider power imbalances. It was also highlighted that it should be taken into account that data is just one representation of reality, and that the same reality may be represented in different ways using different data.

Some participants noted that, while it is true that Global South countries might face challenges around missing, fragmented and incomplete data for policy-making, this also represents an opportunity to build policy-relevant datasets in a way that is bottom-up, taking into account local contexts and needs. However, that also means that scale may be harder to achieve: this challenges the dominant paradigm of AI and machine learning that rely on large-scale data availability to train models. This presents an interesting tension between localised approaches to data collection and use, and the potential strategic value of 'small data', and current dominant 'big data' paradigms in the fields of AI and machine learning.

2 | Rentier innovation 

It was argued that big tech companies are not creating value from data through meaningful innovation, but rather are using traditional tactics for establishing monopoly power and engaging in simple rentier innovation. Rentier innovation - or 'rentierism' - can be understood as the capacity to derive income from controlling access to assets; rentierism boosts profitability by taking advantage of social and political conditions in order to capture value, rather than creating value. It was argued that by relying on underpriced labour and data collection and enclosure, big tech companies overstate their capacity to create value from their data practices, because they are actually innovating ways to extract value from data rather than innovating in ways to create value from data. It was noted that narratives that overestimate the utility of big tech platforms’ data practices can even take the form of criticising their potential negative impacts: ‘tech companies are happy having people label them as ‘evil geniuses’, as long as they are still called ‘geniuses’.’

It was argued that the logic of rentier innovation around data and digital infrastructure both drives the development of new digital products and services, and is also central to visions of development and nation-building around data and digital infrastructure in the Global South. For example, it was noted how policy-makers claim that a country needs to leverage the personal data of its citizens for economic growth and development, and thus the state becomes engaged in developing the digital infrastructure that can enable a continued form of rentier innovation. This critical perspective on value creation from data in the Global South gives us an opportunity to rethink the development of digital infrastructure and the Fourth Industrial Revolution in the Global South. It was noted that when people are asked to exchange data about themselves for access to an essential service, this can hardly be called a meaningful choice. 

3 | Changing paradigms

It was noted that there is a particular way in which we often view personal data versus other kinds of data when discussing value creation from data or opening different datasets; but this distinction goes hand in hand with linking 'ownership' of data to the production of that data - i.e. the idea that whoever produces data or contributes data to an ecosystem also 'owns' that data. It was observed that one possible approach for addressing the issue of rentierism in data economies and digital infrastructure might be to reject this idea that equates production of data with data ownership. An alternative paradigm might be to consider people as being both resources for data extraction and targets of data extraction. For example, it was noted how the way in which models are trained, the data used to train them, and the work of the people on the ground generating that training data (as resources), can be delineated from the way in which models target people to influence their behaviour. In the context of government programmes, for example, one can conceive of government officials and communities participating in the data generating process as resources, and then government officials making decisions about service provision as well as the communities being targeted for access to services as targets. This delineation between people as resources for data practices and people as targets of data practices may be a more useful way of framing the relation between the process of value creation and extraction from data, and the work that people do in that process of data value creation. 

Additionally, it was noted how cheap labour underpins data collection or production for rentierism in data economies and digital infrastructure, and also impacts work practices. For example, it was highlighted how the introduction of automated systems to detect diseases has affected the role of health workers by moving them away from engaging with patients, and instead focusing their work on evaluating the analytical outcomes of the automated system; and how this, in turn, changes the kinds of knowledge and skills that the next generation of health and care practitioners will need.


Roundtable attendees highlighted the need to see data as a collective resource for communities, rather than as private property for individuals or companies. This might mean changing the focus from the data rights of individuals standing in opposition to the interests of companies, to a structure that focuses on respecting and responding to local culture and cooperative ownership.

Listen to the talk here: https://theodi.org/article/octavia-and-data-imaginations/ And add your comments here

Browse categories

Scroll right
illustration by:

Personal Data, Rentier Innovation, and the Global South

Can a model of personal data as an economic asset deliver AI for social good? At a talk at the Open Data Institute on AI and Infrastructures in the global south, Urvashi characterized current innovation paradigms as one of 'rentier innovation'. This paradigm is harmful and unsustainable. Listen to her talk, and see the summary notes from ODI.

The history of current AI and Machine Learning developments was explored to highlight how data practice and governance can avoid exacerbating inequalities in the Global South. By looking at how the interests of the advertising industry have shaped the way data is generated and captured by big tech platforms, claims made about value creation in digital infrastructure were questioned. Instead it was proposed to frame big tech business models as a new form of rentier innovation: this framing of data-driven economic activities as rentiership enables a new critical perspective on the development of the Fourth Industrial Revolution. For example: how do we avoid turning personal data into the currency of our times? How do we ensure people and communities have meaningful choices in regards to how data about them is collected and used?

1 | Less data, more context

It was argued that most challenges and tensions around how to govern data infrastructure in the Global South tend to be about competing values and interests; however, it was also argued that data availability and data in itself contribute to determining these values and interests, or may even be used strategically for shaping these determinations.

Participants discussed how framing policy-making as being data-driven, and assuming that more data will always solve policy questions, can mask issues around the underlying problems that policy needs to address. For example, framing issues around malnutrition amongst poor populations as a problem that could be solved if we had access to the right data may divert attention from other issues that may be driving it, such as corruption and mismanagement. Consequently, the importance of being able to diagnose policy problems is essential for making sure there is clarity and transparency on what the data can and will be useful for, and therefore which data needs to be collected. For this reason, it was argued that a key question to consider is: what kind of policy-making is data extraction in the Global South enabling; and whether the benefits for policy-making of expanding data extraction justify this extractivism or its expansion. 

Participants also noted how over-reliance on data for policy-making shapes how we evaluate and compare the level of development of different territories or jurisdictions in the Global South in ways that may be misleading. It was observed that relying on data without understanding the context on the ground in which that data is generated may cause mismatches between the interpretation of the data by policy-makers and the material or experiential realities in the domain about which policy is being made. This raised questions about who is able to access, provide, use and interpret data for policy-making, and how that ability is intertwined with other sources of power or wider power imbalances. It was also highlighted that it should be taken into account that data is just one representation of reality, and that the same reality may be represented in different ways using different data.

Some participants noted that, while it is true that Global South countries might face challenges around missing, fragmented and incomplete data for policy-making, this also represents an opportunity to build policy-relevant datasets in a way that is bottom-up, taking into account local contexts and needs. However, that also means that scale may be harder to achieve: this challenges the dominant paradigm of AI and machine learning that rely on large-scale data availability to train models. This presents an interesting tension between localised approaches to data collection and use, and the potential strategic value of 'small data', and current dominant 'big data' paradigms in the fields of AI and machine learning.

2 | Rentier innovation 

It was argued that big tech companies are not creating value from data through meaningful innovation, but rather are using traditional tactics for establishing monopoly power and engaging in simple rentier innovation. Rentier innovation - or 'rentierism' - can be understood as the capacity to derive income from controlling access to assets; rentierism boosts profitability by taking advantage of social and political conditions in order to capture value, rather than creating value. It was argued that by relying on underpriced labour and data collection and enclosure, big tech companies overstate their capacity to create value from their data practices, because they are actually innovating ways to extract value from data rather than innovating in ways to create value from data. It was noted that narratives that overestimate the utility of big tech platforms’ data practices can even take the form of criticising their potential negative impacts: ‘tech companies are happy having people label them as ‘evil geniuses’, as long as they are still called ‘geniuses’.’

It was argued that the logic of rentier innovation around data and digital infrastructure both drives the development of new digital products and services, and is also central to visions of development and nation-building around data and digital infrastructure in the Global South. For example, it was noted how policy-makers claim that a country needs to leverage the personal data of its citizens for economic growth and development, and thus the state becomes engaged in developing the digital infrastructure that can enable a continued form of rentier innovation. This critical perspective on value creation from data in the Global South gives us an opportunity to rethink the development of digital infrastructure and the Fourth Industrial Revolution in the Global South. It was noted that when people are asked to exchange data about themselves for access to an essential service, this can hardly be called a meaningful choice. 

3 | Changing paradigms

It was noted that there is a particular way in which we often view personal data versus other kinds of data when discussing value creation from data or opening different datasets; but this distinction goes hand in hand with linking 'ownership' of data to the production of that data - i.e. the idea that whoever produces data or contributes data to an ecosystem also 'owns' that data. It was observed that one possible approach for addressing the issue of rentierism in data economies and digital infrastructure might be to reject this idea that equates production of data with data ownership. An alternative paradigm might be to consider people as being both resources for data extraction and targets of data extraction. For example, it was noted how the way in which models are trained, the data used to train them, and the work of the people on the ground generating that training data (as resources), can be delineated from the way in which models target people to influence their behaviour. In the context of government programmes, for example, one can conceive of government officials and communities participating in the data generating process as resources, and then government officials making decisions about service provision as well as the communities being targeted for access to services as targets. This delineation between people as resources for data practices and people as targets of data practices may be a more useful way of framing the relation between the process of value creation and extraction from data, and the work that people do in that process of data value creation. 

Additionally, it was noted how cheap labour underpins data collection or production for rentierism in data economies and digital infrastructure, and also impacts work practices. For example, it was highlighted how the introduction of automated systems to detect diseases has affected the role of health workers by moving them away from engaging with patients, and instead focusing their work on evaluating the analytical outcomes of the automated system; and how this, in turn, changes the kinds of knowledge and skills that the next generation of health and care practitioners will need.


Roundtable attendees highlighted the need to see data as a collective resource for communities, rather than as private property for individuals or companies. This might mean changing the focus from the data rights of individuals standing in opposition to the interests of companies, to a structure that focuses on respecting and responding to local culture and cooperative ownership.

Listen to the talk here: https://theodi.org/article/octavia-and-data-imaginations/ And add your comments here

Browse categories

Scroll right